Plasticity
The five water levels in the greenhouse significantly affected all traits (Table 1, 2; Appendix Fig. S1), indicating plasticity in response to water. Significant site × water interactions for all traits except reproductive allocation (Table 2) indicated genetic variation in plastic responses among the four sites of the rainfall gradient.
For the two sites with climate manipulations, the magnitude of plasticity (quantified as CV) differed substantially among traits. CV was lowest in δ 13C (3-4%) and days to flowering (c. 10%), intermediate for diaspore weight, height, stomata density, leaf number at flowering and total biomass, and particularly high in reproductive allocation (c. 80%) and seed number (c. 100%) (Fig. 3). The three traits showing rapid evolution thus had low (days to flowering), intermediate (leaf number at flowering) and high plasticity (reproductive allocation), respectively, signifying that plasticity and the probability of rapid evolution in the field were independent.
Climate manipulations had overall little effect on the magnitude of plasticity. Increased plasticity (CV) in plants from dry manipulated plots compared to controls was found in diaspore weight (p=0.01) and a similar non-significant tendency in days to flowering (p=0.06) (Table 3, Fig. 3). Plasticity in vegetative biomass increased in dry plots (p=0.03), but only compared to wet plots (Table 3, Fig. 3). The lack of any significant climate manipulation × site interaction showed that plasticity responded similarly to manipulations in both sites (Fig. 3, Table 3). Moreover, plasticity (CV) was significantly higher in the Mediterranean than the semi-arid site in four traits: days to flowering, height, reproductive allocation, seed number (Table 3, Fig. 3). All plasticity results were robust when using another common plasticity index, PIv (Valladares et al. 2006).